Coding Harness Matters More Than Model: Token Use Varies 9.2x

Tests using the same prompt across frameworks showed token consumption varying up to 9.2x with the same model, suggesting the agent harness matters more than the model itself for cost and latency.

2026-09-08 ~ 2026-09-08 · 2 related posts

• Same model, 9.2x token spread: harness matters more than the model — zainhas · 2026-09-08
• Same model, 9.2x token spread: harness tests show coding agent harness matters more than the model — zainhas · 2026-09-08